the quantitative analysis of glycine in these samples was carried out. These results are also shown in Table 111. The glycine contents in these samples were also determined with an error of within 3%. DISCUSSION
Dioxane was used to dissolve TT since T T is stable in dioxane. The increase in the concentration of dioxane in the reaction mixture accelerated the rate of color reaction. But a high content of dioxane in a 0.2M phosphate buffer solution at pH 8.0 resulted in precipitating phosphate and preventing the occurrence of the color reaction. The concentration of dioxane in the reaction mixture was not over 30%. The reaction mixture became turbid when TT in dioxane was added to the buffered solution at pH 8.0 but the mixture became clear as the reaction proceeded. So it was possible to measure the absorbance of the reaction mixture. In addition to N-substituted derivatives of glycine, N-substituted derivatives of serine, threonine, and tryptophan were also positive in color reaction. Only L-type amino acids and their derivatives were used in this paper, but the results obtained from DL-type amino acids and their derivatives were the same as those obtained from the L-type. Among N-substituted derivatives of glycine, N-benzoylglycine was chosen because the molar absorptivity of the color product from N-benzoylglycine and TT was the largest among the N-substituted derivatives of the glycine tested. Also, the benzoylation procedure was relatively simple.
This paper shows that N-benzoyltryptophan, when treated with the HzOz-dioxane method, was changed to a substance which was negative in color reaction. T o determine the glycine content in proteins, however, it was necessary to hydrolyze the protein prior to the procedure in this paper. Tryptophans in proteins were decomposed by the acid hydrolysis, so that it was not necessary to use the elimination procedure. Such organic compounds as acetylacetone, diethylmalonate, and citric acid were also positive in this color reaction. These have an active methylene group in the molecule. N-Acetylglycine ethyl ester, however, has also an active methylene group in the molecule, but this was not positive in the color reaction. It could not be concluded that a substance which had active methylene groups in the molecule was always positive in this color reaction. Three milliliters of the colored solution, the absorbance of which was at least 0.1, was required for the measurement of the glycine content by the method described in this paper. The molar absorptivity of the product obtained from Nbenzoylglycine and TT was 3.62 X lo4. Therefore, it may be possible to determine the glycine content in amino acid mixtures quantitatively if there is more than 0.63 pg of glycine in the mixture.
RECEIVED for review June 2, 1969. Accepted September 29, 1969.
Determination of Carbon in Thin Films on Steel Surfaces William R. Lee and Lynn L. Lewis Chemistry Department, Research Laboratories, General Motors Corporation, Warren, Mich. 48090 THEIMPORTANCE of the surface characteristics of materials is widely recognized. In our laboratories, for example, steel surfaces have been studied extensively because of their importance in steel-lubricant and steel-paint systems. Analytical techniques developed for characterizing steel surfaces have been used to provide quantitative data routinely for microgram quantities of sulfur, chlorine, phosphorus, and zinc on ball bearings ( I ) , and additional elements on sheet steel. No technique was available, however, for determining carbon in thin carbonaceous films of varying thickness that originate from oil, grease, etc. The determination of carbon on these surfaces presents a unique analytical challenge. Carbon is also present on and below the steel surface as carbon in solution and as metal carbides, and there is no means of directly measuring the carbon of organic origin. Therefore, the essential feature of an analytical technique is a means for selectively removing the carbon of organic origin from the surface prior to the measurement so that there will be no interference from the carbon in the steel. Previous work on determining the amount of organic residues on metal surfaces has been limited to two studies, both of which are based on heating the sample in oxygen. Combustion conditions were chosen to provide complete oxidation and removal of the organic residue, with the combustion temperature being relatively low to avoid removal of carbon (1) J. L. Johnson, Microchem. J., 8,59-68 (1964).
from the steel. Boggs and Pellissier ( 2 ) heated sheet steel at 500°C in oxygen at low pressure for 10 minutes and measured the carbon dioxide produced using a manometric technique. Solet (3) reported that carbon on the surface of nickel strip could be determined by heating the sample in oxygen at 600 "C and measuring the change of electrical conductivity of a solution whichabsorbed the carbondioxide that was formed. These combustion methods are straightforward and of adequate sensitivity, but they cannot be applied directly to steels covered with films of unknown origin; the proper conditions of time and temperature must be established beforehand for each particular sample. The following equatons illustrate the chemical reactions that are important in the combustion method: C(organie)
+
'12 0 2
= CO
(1)
+ CO M[C] + = MO + CO MO + MC = 2M + CO MC+
=
0 2
MO
(2) (3)
0 2
co +
'12 0 2
=
(4)
coz
(5)
where C(arganio) is a carbonaceous film of organic material, MC is a metal carbide, M[C] is carbon in solution in the metal, (2) W. E. Boggs and G. E. Pellissier, Muter. Res. Std., 1, 627-630 (1961). (3) I. S. Solet, ANAL.CHEM., 38, 504 (1966).
ANALYTICAL CHEMISTRY, VOL. 42, NO. 1, JANUARY 1970
a
103
Gas Chromlatoyraph
1.0
z d c 0
a m
p_ _ _ _ _ PH1
L - -
A
L 01
k
I
0.5
U
-----I
c
n
Figure 1. Schematic diagram of equipment for determining carbon on surface of metals PR1 Pressure regulator PG Pressure gauge F M Flow meter NV Needle valve GSV Gas sampling valve, Perkin-Elmer Corp., Model 154-0068 PR, Pressure regulator, Moore Products Co., Model 40-2 IP Injection port T Manganese dioxide trap Flow regulator, Watts Regulator Co., Model M1, Type 263 FR C Column of silica gel, 20-cmlong
and MO and MC are a metal carbide and oxide that take part in deoxidation reaction when the metal is heated. Any carbon monoxide present is oxidized, Equation 5. In general, the chemical reaction given in Equation 1 can be expected t o occur at lower temperatures than the others. The rates of the reactions given in Equations 2-4 will be limited by diffusion, particularly for the deoxidation of the metal, illustrated in Equation 4. Thus, it appears that a chemical separation can be based on heating the sample from ambient to a higher temperature in a stream of oxygen, with the carbon of organic origin being oxidized and removed first. For three important reasons the analytical method should provide for the continuous measurement of the carbon dioxide as it is formed. First, the shape of the carbon dioxide evolution curve is useful in the interpretation of results because organic films will burn off a t different temperatures, depending on the source and the history of the film. Second, some means is needed for determining when combustion of the organic film is complete. Third, a means for monitoring carbon dioxide evolution is necessary for an evaluation of the blank contributions arising from the reactions given in Equations 2-4. Previous work on determining small quantities of carbon in metals by combustion techniques indicated that thermal conductivity measurement techniques are sensitive enough to measure traces of carbon dioxide continuously in the oxygen stream (4). Therefore, we chose that technique for our measurements. EXPERIMENTAL Instrumentation. A diagram of the equipment is shown in Figure 1 ; the components of the gas chromatograph (PerkinElmer Corp., Model 154) appear in the blocked in area. Oxygen flows through the gas chromatograph and a gas sampling valve (GSV), which is used to by-pass the combustion equipment when specimens are added or removed. A pressure regulator, PR2, maintains a constant pressure (2 psig) in the combustion tube when it is heated. For calibration purposes, quantities of carbon dioxide are added to the oxygen stream through the injection port (IP) with a gas syringe. Specimens are placed in a quartz tube, 1-inch (4) L. L. Lewis and M. J. Nardozzi, ANAL.CHEM., p 1214. 104
TempEratUrE, "C
Figure 2. Detector response to carbon dioxide produced when ball bearings exposed to oil additives are heated in a stream of oxygen A. Chlorinated wax
B. Sulfurized terpene oil C. Didodecyl phosphite
in diameter, that is held in the combustion furnace; the resistance furnace (Hevi-Duty Electric Co., Type 70-T) is the hinged, tube-type with a 12-inch heating zone. The gases from the combustion furnace pass, in series, through a catalyst furnace (900 " C ) containing silver vanadate and platinized copper oxide for the complete conversion of combustion products to carbon dioxide, a manganese dioxide trap t o collect any sulfur dioxide, a flow regulator (FR) that controls oxygen flow at 60 ml per minute, a silica gel column (C) that absorbs water formed in the combustion, and the measuring side of the thermal conductivity detector. The response of the detector is recorded on a strip chart recorder (0-1 mV), with the peak area being a measure of the carbon doxide that is formed. Temperatures are measured with a thermocouple positioned near the specimen inside the combustion tube. Approximately 90 seconds elapse before the carbon from the organic surface film is swept through the detector as carbon dioxide; the temperatures cited later are corrected for this delay. Analytical Procedure. A daily calibration is based on peak areas measured with a planimeter. The relationship between the carbon dioxide added and the measured peak area was linear over the range of interest, 0 to 200 pg of carbon. (The recovery of carbon from known amounts of stearic acid, potassium acid phthalate, and mineral oil confirmed the validity of the calibration procedure.) For the determinations, specimens are placed in the combustion tube and centered in the furnace. Ball bearings are placed in a pre-ignited, glazed porcelain boat; sheet specimens (measuring 0.5 by 4 inches) are laid in the tube. With the specimen in place, the oxygen line is connected t o the inlet of the combustion tube, and the gas sampling valve is positioned so that oxygen passes through the combustion tube. When the system has stabilized, as indicated by a straight base line on the recorder, the furnace is turned on (110 V). (Specimens are heated reproducibly at a rate determined by the characteristics of the furnace.) When the combustion is complete, as indicated by the return of the recorded curve t o the base line, the furnace is turned off and cooled for the next determination. (Rapid cooling is accomplished by raising the hinged section of the furnace and directing a jet of air across the combustion tube.) Approximately 20 minutes are required for the combustion, and 15 minutes t o cool the furnace for the next determination. The base line drifted only during the first two minutes of the heating cycle, which preceded the recording of the peak.
ANALYTICAL CHEMISTRY, VOL. 42, NO. 1, JANUARY 1970
Time min
200
500
400
Table I. Precision Study (Stearic Acid on Steel) Standard Coefficient 12 deviation of variation Carbon, p g Within day: 36.5 8 1.13 3.10 141.5 8 1.70 1.20 38.6 7 1.11 2.88 Day-to-day: 141.6 7 2.27 1.60
20
15
10
500
600
Temperature, 'C
Figure 3. Detector response to carbon dioxide produced when sheet steel treated with oil is heated in a stream of oxygen A . Freshly prepared sample B. Aged samples
Commercial grade oxygen is used without any purification. Impurities in the oxygen do not interfere because a continuous gas flow system is used with the oxygen passing through both sides of the detector. RESULTS AND DISCUSSION
Precision Studies. To test the method for both withinday and day-to-day repeatability, two series of specimens were prepared by adding known amounts of stearic acid to steel coupons that had been oxidized beforehand at 600 "C to remove surface carbon. (The stearic acid was dissolved in benzene and delivered precisely from a 50-p1 buret, after which the benzene was evaporated.) The results of this precision study are shown in Table I. The first series of coupons was designed to test the method at the 30- to 40-pg level of carbon, which is the normal lower range of application of the method. The second series was designed to test the method at the 140- to 150-pg level of carbon, which is the normal highest level of application. T o provide the within-day data, the test design called for calibrating the instrument by using the carbon dioxide injection technique and determining the carbon on each of eight coupons within a single day at each level. For each of the next seven days, the instrument was calibrated with carbon dioxide injections, and a single determination was made at each level. For all determinations, the average of three planimeter measurements was taken as the area of a peak. As shown, the coefficient of variation at both levels is quite acceptable on a daily and a day-to-day basis, with no statistically significant difference between the two. Recovery was 9 8 . 2 z at both levels. Carbon on Ball Bearings. The technique has been applied to a study of the films formed on ball bearings during friction tests made with various oils and oil additives at different temperatures (5). Curves obtained with ball bearings exposed to three different oil additives (in the same base oil) are shown in Figure 2. The surface films contain carbon as well as some of the inorganic elements (P, S, and Cl) that are present in the additives. Most of the ball bearings have between 10 and 90 pg of carbon per square inch of surface. (A significant fraction of the carbon on the surface is derived apparently from the base oil rather than the additive.) ~
~~~~
( 5 ) F. G. Rounds, private communication, Research Laboratories,
General Motors Corporation, Warren, Mich., February 1969.
Table 11. Variation of Carbon on Surface of Adjacent Specimens Taken from Steel Coils Surface carbon after cleaning. ugisa in. TrichloroAbrasive ethylene Coils Specimens vapor spray A 1 5 17 2 23 4 5 3 19 B 1 14 5 2 5 13 3 12 5 C 1 18 4 2 22 3 3 16 3 D 1 29 5 2 30 4 3 27 4
Carbon on Sheet Steel. Another application of the method concerned determining carbon on the surface of sheet steel for an investigation of the relationship between surface composition and corrosion susceptibility. As shown in Figure 3, the burning characteristics of the substances in the surface film vary, as does the quantity of carbon on the surface. The carbon dioxide from the specimen that had been treated recently with rolling oil was formed in a lower temperature range than that obtained with aged specimens taken from steel coils. The oil on the coils had apparently formed a varnish-like film of higher oxidative stability. The surface carbon values obtained on sheet steel specimens are usually between 5 and 40 pg of carbon per square inch. Organic residues are present also on the steel after treatment with a phosphate primer coat, as shown in Figure 4. The irregular shape of these curves is attributed to the presence of substances that burn off at different temperatures. As stated earlier, analytical conditions must be selected that have been established to oxidize selectively the carbon of organic origin on the surface, with no appreciable blank from the carbon in the steel. These curves show that carbon dioxide formation is no longer detected at temperatures of about 600 to 650 "C. More carbon dioxide is evolved upon further heating, however, beginning at approximately 700 "C; the carbon removed above 700 OC increased sharply with increasing temperature and undoubtedly originates in the steel. Variation of Carbon on Sheet Steel. Surface carbon at different locations within a steel coil may vary considerably (2). Replicate determinations should therefore be made to establish the variation that will occur on a specimen-to-specimen basis. The results obtained on adjacent pre-cleaned specimens taken from four different coils are shown in Table 11. The measured variation among the specimens within a coil includes a variation of unknown significance arising from the cleaning procedure.
ANALYTICAL CHEMISTRY, VOL. 42, NO. 1, JANUARY 1970
0
105
Time, rnin
Table 111. Variation of Carbon on Surface of Steel Panels after Various Cleaning Treatments Surface carbon, pg/sq in. Cleaning treatment Panel 1 Panel 2 Panel 3 Trichloroethylene vapor 5 15 30 Lacquer thinner 3 10 35 Acetone-toluene-chloroform (1:l:l) 3 3 40 Hydrochloric acid, 50 3 13 11 Abrasive jet spray 2 2 10
10
1.0
Table IV. Carbon on Surface of Different Steels after Cleaning Carbon Surface carbon, pg/Sq. in. content, %, Solvent Abrasive Steel nominal cleaning spray 1018, low alloy 0.18 20 6 4118, low alloy 0.20 14 52100, low alloy 1 .o 13 5 M50, high temperature 0.8 27 ... 440C, stainless 1.1 21 9 301, stainless 0.1 16 8 I
.
20
i
Ternperatura, 'C
Figure 4. Detector response to carbon dioxide produced when phosphated steel panels are heated in a stream of oxygen
.
Some specimens (all were Type 1018 steel) were pre-cleaned with trichloroethylene vapor to remove residual oil. Others were pre-cleaned with a n abrasive jet spray of alumina and water. Results obtained with the abrasive treatment are of particular interest because it is believed that they represent the magnitude of the inherent blank, which is due to the carbon in the steel, This type of abrasive cleaning was chosen for this comparison because organic films on the surfaces are removed (with some steel), and the surface is not heated, as in dry grinding. The quantity of carbon found on the samples cleaned by the abrasive spray agrees closely with that reported elsewhere, in' which approximately 6 pg of carbon per square inch was obtained on sheet steel that was pre-cleaned electrolytically ( 2 ) . Specimen Pre-Cleaning. The pre-cleaning procedure to be used to remove oil from specimens prior to the determination of surface carbon should be selected early in a n experimental program. As shown in Table 111, results may vary when different treatments are used. The effectiveness of a particular cleaning treatment will depend, of course, on the treatment time and the nature of the substances on the surface. In addition, these data show that treatments with hydrochloric acid and the abrasive spray were not equally effective in removing carbon from the surface. Effect of Type of Steel. Another variable is the effect of the type of steel and its carbon content on the quantity of carbon removed in the combustion process. The data in Table IV were obtained on ball bearings made of various types of steel that contained from 0.1 to 1.1% carbon. The bearings were pre-cleaned with a solvent or the abrasive spray. Assuming that the abrasive spray cleaning provides a surface free of organic substances and does not selectively remove carbon from the steel, these data indicate that the type of steel and its carbon content is not important. [Large
106
15
I
A , B. Commercial sheet coated with different phosphate primers
differences in the quantity of carbon found on surfaces cleaned with the abrasive spray were expected because these steels oxidize at different rates, and they contain metal carbides that ignite at temperatures below 600 "C (6).] Application of Technique. No apparatus blank is detected when the procedure is followed without a specimen. Results obtained on various steels treated with the abrasive spray indicate that approximately 6 pg of carbon per square inch of surface are removed from the steel itself in the combustion. This quantity is significantly less than that present in surface films, (The results presented were not corrected for this carbon contribution from the steel.) The limit of detection is 8 pg of carbon. Precise results have been obtained on trace quantities of carbon present on metal surfaces that have been subjected t o a variety of conditions. Small changes in the carbon content of surface films are detected readily with this equipment, which can be assembled from items of low cost that are common to a n analytical laboratory. Analysis time is one-half hour per specimen, Information obtained with this technique has filled a definite need for more information about metal surfaces, This information cannot be obtained conveniently by other means. ACKNOWLEDGMENT
The authors thank M. D. Cooper, J. L. Johnson, and A. C. Ottolini for helpful discussions during the course of this work. RECEIVED for review July 28, 1969. Accepted October 17, 1969. Presented before the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, March 1969. (6) H. S. Karp, W. R. Bandi, and L. M. Melnick, Tulmzta, 13, 1679-1687 (1966).
ANALYTICAL CHEMISTRY, VOL. 42, NO. 1, JANUARY 1970
